Re: [kernel-hardening] [PATCH 0/6] move mmap_area and PIE binaries away from the stack
On Fri, 2017-06-02 at 21:37 -0700, Kees Cook wrote: > > Patches 3 and 4 need some (minor?) adjustments It's currently a bug fix. Doing something further would be more than fixing the bug and should probably be separate.
Re: [kernel-hardening] [PATCH 0/6] move mmap_area and PIE binaries away from the stack
On Fri, Jun 2, 2017 at 8:20 AM, wrote: > There are a few bugs causing the kernel to sometimes map PIE > binaries and the mmap_area where the stack is supposed to go. > > This series fixes them for x86, ARM64, and PPC. > S390 seems to be ok. > > If people are fine with this approach, I can work my way > through other architectures, too. Andrew, I'd rather the ELF_ET_DYN_BASE changes not be in -mm yet. I think it's the wrong approach. Patch 1 (comment update) is fine. Patches 3 and 4 need some (minor?) adjustments, but the ELF_DYN_BASE patches need much more thought, IMO (I've sent a separate email about those.) And please add me to explicit the Cc list for this series. -Kees -- Kees Cook Pixel Security
[PATCH 0/6] move mmap_area and PIE binaries away from the stack
There are a few bugs causing the kernel to sometimes map PIE binaries and the mmap_area where the stack is supposed to go. This series fixes them for x86, ARM64, and PPC. S390 seems to be ok. If people are fine with this approach, I can work my way through other architectures, too.
